Antitrust regulators zero in on AI

The watchful eyes of US antitrust enforcers are squarely on the artificial intelligence industry.

Last week, the US Federal Trade Commission announced it was opening an inquiry into multibillion-dollar investments by tech giants into smaller AI startups. Amazon, Google, and Microsoft made investments in Anthropic and OpenAI, and while they didn’t buy them outright, that has not stopped federal antitrust regulators from flexing some muscle.

Microsoft poured $13 billion into OpenAI, the company that ushered in the start of the AI boom with the release of its chatbot ChatGPT in November 2022, and the FTC is also probing two separate investments into Anthropic, which makes the AI-powered chatbot Claude, by Amazon ($4 billion) and Google ($2 billion).

It’s possible that in a more hands-off regulatory environment, these Silicon Valley stalwarts would have simply bought the pure-play startups outright. But doing so these days would enlarge the targets already on their chests.

The US government’s commitment to busting corporate dealmaking in the internet sector has been spotty over the past two decades. The historical rate at which the government challenges mergers is “far, far lower in the digital sector,” says Diana Moss, vice president and director of competition policy at the Progressive Policy Institute. This is research she oversaw and testified about to Congress in her previous role as the president of the American Antitrust Institute.

Federal antitrust enforcement is now led by FTC chair Lina Khan, a longtime critic of Big Tech dating back to her days as a student at Yale Law School, and the Department of Justice’s antitrust chief Jonathan Kanter, who spent his final years in private practice in part representing smaller tech firms in lawsuits against Apple and Google. In the first few years of their tag-team tenure, Khan and Kanter have sued Google for abusing its monopoly in advertising, sued Amazon for anticompetitive behavior in the online retail market, and unsuccessfully sued Meta to block its acquisition of the VR firm Within. Khan scored a big win in December when a federal court upheld the agency’s decision to block a $7.1 billion biotech merger, and several tech companies including Adobe and Figma have terminated merger plans after meeting with antitrust regulators. Still, it could take years for Khan and Kanter to notch their first major victory over Big Tech.

In a recent speech at Stanford University, Khan said the government wouldn’t turn a blind eye to anti-competitive dealmaking in the AI space, noting that the FTC “will be clear-eyed in ensuring that claims of innovation are not used as cover for lawbreaking.”

Brian Albrecht, chief economist for the International Center for Law & Economics, said there’s no question that Khan “believes there was too little scrutiny on previous tech acquisitions and wants to get ahead.” He says she’s been overeager with a “desire to bring any tech case, instead of good cases” (such as the Meta-Within case). Still, while the FTC hasn’t yet brought a case against these AI investments, Albrecht says it “has a flavor of ‘we need to do something, and this is something.’”

The FTC inquiry is just that — merely an inquiry. The agency hasn’t yet launched a formal investigation into any of these deals, which would be a necessary step before it decides whether to bring lawsuits. In fact, recent reports indicate that the FTC and DOJ both want to investigate Microsoft’s stake in OpenAI but can’t agree over who’ll get to do it.

But it’s a warning shot, a declaration of intent, a resolution that the investment-not-acquisition strategy — if that’s the strategy after all — will not go unnoticed.

Investments, not acquisitions

Antitrust regulators have broad authority over partial-ownership investments, not just full-on corporate takeovers. That’s important, Moss says, because her research shows that the percentage of investments in AI over the past three decades is about three times higher than that of acquisitions involving AI. “That tells you a lot about how companies are approaching AI,” she says.

Microsoft’s arrangement with OpenAI is somewhat stranger than the others because while it’s invested an astronomical sum in the ChatGPT maker, OpenAI is technically run by a nonprofit. Until recently, Microsoft didn’t even occupy a seat on that nonprofit’s board! But when the board dismissed OpenAI CEO Sam Altman in November, Microsoft’s power was hard to ignore. Microsoft promised to hire all of the 700 employees threatening to leave OpenAI over the ouster, successfully lobbied for Altman’s reinstallation, and won a (nonvoting) board seat in the aftermath.

“The arrangement does not get some sort of special immunity because it isn't a standard investment,” Albrecht says. “That being said, investments, joint ventures, strategic partnerships have often (and should) received more leniency from the agencies.”

And even though OpenAI is run by a nonprofit, that doesn’t obviate the need for antitrust enforcement. “The exercise of market power affects prices, quality, and innovation similarly in the case of for-profit and nonprofit organizations,” Moss says, noting that many universities and hospitals have nonprofit status and have received antitrust scrutiny.

The UK’s Competition and Markets Authority is already investigating Microsoft’s investment in OpenAI, and Microsoft has defended itself by pointing to the odd nature of its investment. Instead of buying equity in OpenAI, Microsoft receives half of the startup’s revenues until the $13 billion investment is repaid, according to the Los Angeles Times.

A new era for antitrust

In the past few decades, Silicon Valley technology companies have become the most valuable firms in the world. Seven of the top nine most valuable firms in the world are tech companies with AI investments (Amazon, Apple, Google, Meta, Microsoft) or chip manufacturers (NVIDIA and TSMC), all of which have massive direct or indirect interests in the success of AI.

Many critics of these Big Tech firms say they have grown bloated and unruly without proper antitrust enforcement to keep them from gobbling up competitors. That seems to be the view of Khan and Kanter, too — plus, many overseas antitrust regulators who could make life uncomfortable for any of these global companies.

And these companies know that.

It’s hard to know whether in another time, facing different scrutiny, Microsoft might have tried to buy OpenAI. Or if Amazon or Google would’ve made an offer to buy Anthropic.

“The current state is that any Big Tech company has to worry about the FTC for any major investment or business decision they make,” Albrecht says. “That makes investments relatively more attractive than acquisitions.”

But this inquiry might reveal that the gap, he says, isn’t as big as the companies in question — some of the biggest AI firms in the world — might wish.